Using Information about Influencing Factors to Split Data Samples in Machine Learning Methods for the Purposes of Assessing Information Security

Improving quality indicators of determining information security states of individual segments of cyber-physical systems involves processing large data arrays. This article proposes a method of splitting data samples to improve the quality of algorithms for classifying information security states. Classification models are configured on training sets of examples that may contain outliers, noisy data, and imbalances of observation objects, which affects the quality indicators of the results. At certain points in time, the influence of the external environment may lead to changes in the frequency of observed events or the ranges of logged values, which significantly affects quality indicators. It has been shown that a number of events in samples are caused by the influence of internal and external factors.